Wow! I am very surprised to hear this given how long I have used Julia, and how much I prefer it to Python.
The reason for our indexes starting at 1 is that it is quite common in languages like Python to write an algorithm and then have to subtract 1 from indexes all over the place because; in math, 0 means 0, 0 doesn't mean one. Therefore, Julia's " just like the papers" approach would pretty much require that indexes start at 1.
It is upsetting to hear that you did not like Julia though, as I wish other Scientists could get the opportunity to feel the same about the language that I do.
The power of Julia doesn't come when you use Python packages. Something like that was much more necessary a few years ago, obviously calling C code from Python code from Julia is slower than calling C code from Python... So I agree with you on your stance of " why switch to Julia to use Python packages?" But I also can assure you, this is not what most Julia users are doing.
Julia is great because when you start a new project in it, your projects only limitation brought by the language is the ecosystem of the language; if you are willing to create, then you will not face any problems.
If you wanted to implement, or prototype a new model in Python, it would be very hard, or impossible, to get by creating the model without involving C.
I believe Python is a great language, but I think Python is only a " good" language for Data-Science because for that application it requires the use of an entire other language.
In my past experience, when I worked on teams with Python, we would very often run into the limits of the language in a ton of different aspects. I was the only member of this team who was any good at writing C, thus I was always the one to take care of it when these issues arose.
When it comes to data analysis, and modeling, they are both significantly easier to do in one of these high-level languages, like Julia.
Julia seems like a natural evolution for general-purpose programming from Python.
It needs to also be noted that the languages are similar in some ways, entirely different in other ways. Julia is even more type explicit, and the typing is very strong. Don't go into Julia thinking it is going to be Python, as the two languages reside in completely different paradigms.
I would encourage you to try again with Julia, because there is a pretty good chance of an industrial shift towards the language in the coming years. With R, the shift to Python happened similarly where Universities began picking it up and then 5 years later: boom!
Have you ever trained a random forest classifier off of a train set of 3,000,000? Now imagine that you can do that, but you don't need a model written in C, only Julia.
Julia also has multiple dispatch, and syntax rules that in my subjective views of the syntaxes to both languages, completely blow Python out of the water.
Multiple dispatch also allows you to import and extend any method.
Really, I am somewhat amazed to hear that someone tried Julia and then preferred Python afterwards, that being said I am curious how much you have actually used Julia, as I could see one coming in and seeing the weirdness and being put-off. I remember constantly thinking things were odd syntactically when I entered the language. Now it is second to none for me, and I can hardly stand to use Python because of a combination of whitespace as syntax, slowness, lack of extendability, useless annotations to arguments that do nothing, lack of method errors, the common need for C, implicit inheritance, lack of meta-capability, lack of support for large signed integers, and the inability to work with large data-sets. A lot of these are personal, subjective gripes, but a lot of these gripes are very objective and one can see the benefit to using the language from them.
Just some of my thoughts! I hope you consider giving the language a try, as it is something I have fallen in love with myself!